Research On The Decision-making System Of Virtual Human Football

Virtual human football is a concentrative embodiment of virtual human technology and artificial intelligence, of whose own is a typical multi-agents system and supplies a standard platform for checking and evaluating its theory research. In dynamic virtual match environment, virtual players accomplish defense or attack aim together by combining personal professional skill and collaborating each other. On introducing present research situation and analyzing characteristics of virtual human football and on the basis of predecessors' research achievements, this paper includes mainly the following three works.Firstly, On references of research achievement of robot football decision-making system, this paper introduces a decision generating approach that using both global decision-base and individual decision-base, builds up a layered structure model of decision-making system of virtual human football, which makes players have both the ability of self-determination and overall team strategy simultaneously, thus be more closer to the actual human football game.Secondly, on research of behavior choosing strategy of dominative player, this paper uses artificial potential field method to model the environment that virtual players are in, introduces an algorithm of dominative player's behavior decision-making by using a modified ant colony algorithm. Further more, the methods of searching pass path and self-study process of dominative player also been discussed simply.Thirdly, a decision-making system software is designed and implemented which is divided into different modules according to different functions, which separates decision generating and game data computing and displaying, thus makes the decision-making software has a strong expansibility and flexibility, and within it the validity of the algorithm of dominative player's behavior decision-making is proved.